Epidemiology Flashcards
what is the difference between cohort, case control and cross sectional studies?
cohort studies are observational studies where the patient has EXPOSURE first, then we measure outcome
case-control is when the outcome is known first, then we elucidate exposure
Cross-sectional is when both outcome and exposure are measured simultaneously
what are the major issues with case control studies?
recall bias in the exposure group
selection bias can occur in the control group
what are the major issues with cohort studies?
may suffer from selection bias
cost and time, and long follow up is required
the issue with long f/u is the potential loss to follow up
what is the best style of study for testing prevalence or frequency?
a cross-sectional study (single time point observational study)
what is the formula for positive likelihood ratio?
+ test if disease present/ + test if disease absent
this works out as Sn/1-Sp
what is the formula for negative likelihood ratio?
-ve test if disease present/-ve test if disease absent
this can be worked out as:
1-Sn/Sp
these are <1
what is the rule of 15% with LR?
the rule is about the max change in post test prob that we can see with a LR
if LR+ = 10, or LR- = 0.1, change is up to 45%
LR+ = 5, or LR- = 0.2, change is up to 30%
LR+ = 2, or LR- = 0.5, change is up to 15%
ie, pre test prob is 20%, then LR+ is 10, it can get to about 65%
or pre test is 50%, and LR- is 0.1, it can change to 5%
what is a R-O-C curve?
this is a receiver operating curve
it is a way of plotting true positives versus false positives.
it’s a useful way to determine a cut off whereby we maximise sensitivity without negatively impacting specificity
a handy way to compare diagnostic tests is to compare the area under the curves for each test. (the perfect test would have an area of 1.0. The closer to 1, the better
55.47
46.40
